4 ms·

Measures are not falsified. Hypothesis are (or should be). When you have a hypothesis, or a theory, that can't be deemed false by new data or new experiments, y

by bprieto 12y ago

Measures are not falsified. Hypothesis are (or should be). When you have a hypothesis, or a theory, that can't be deemed false by new data or new experiments, you don't have science. And this is the problem with those "a posteriori" explanations of the hiatus. They are saying that the models that predict a global warming catastrophe can't be wrong, instead of recognizing that the models lack predictive power and they must be adjusted. But if the models can't be wrong they are not science.

anon1385 12y ago
>They are saying that the models that predict a global warming catastrophe can't be wrong Who said that? The incompleteness of climate models isn't something that has just come up recently - they have always been incomplete (they can't be anything but - 'all models are wrong') and that is well understood by anybody who develops or uses them. There are large sub-systems that global circulation models don't even really attempt to model at this stage. Others where they are widely recognised to do a very bad job (e.g. ice sheets). Others that are just modelled stochastically because we don't know how to predict those processes on a year to year basis (e.g. ENSO). Finding inconsistencies between the models and reality is one of the ways that they are improved. I don't know where you are getting these ideas about how climate modellers think, but they don't bare any relation to how climate models are actually developed.
- bprieto 12y ago
  OK, it was a figure of speech. They are not saying that. They are acting like that. The climate scientist who are trying to find the "missing heat" instead of reconsidering their models and revising their predictions of warming.
  - anon1385 12y ago
    Why do you think 'find the missing heat' is different to 'reconsidering their models'? If ocean circulation is transporting more heat into the deep ocean than we thought, then that is something to change in the models. If you are asking, why have they not thrown out our entire understanding of the climate system and started over, then I would argue that isn't a way to actually improve our understanding of the climate. We would be starting over every day. Out understanding clearly isn't all wrong. There are all sorts of other lines of evidence used to develop our understanding of the climate system. From things that are reproducible with simple experiments (or just application of basic physical laws), to things that have taken large investments and long term study to work out. There can be all sorts of reasons that the model got it wrong (assuming it did, I don't think we really have enough evidence one way or the other yet for the idea proposed in the OP - it seems like a bit of a curve fitting exercise at the moment). Not only is our understanding of the systems incomplete, our measurements are incomplete as well (and that gets much worse even just a few years ago - we've only had the Argo network of floats taking ocean temperature for a bit over a decade for example). Often there is just as much reason to suspect that it's our measurements that are wrong than that it's our understanding of the system. It's just the nature of working on a system the size of the entire planet. It's not easy to measure everything, and ever harder to measure what was happening in the past. There have been real issues where it was the data that was wrong and not the models. See the UAH troposphere temperature record for example: https://en.wikipedia.org/wiki/UAH_satellite_temperature_dataset#Corrections_made https://en.wikipedia.org/wiki/UAH_satellite_temperature_data... It contradicted other data sets and the models, and it turns out there were errors in the way the temperatures were calculated. That is a record run by scientists very much on the skeptic end of the spectrum by the way, if you were getting worried about any kind of conspiracy.
    - cicero 12y ago
      I think you give some good explanation of why climate modeling is very difficult. However, the mainstream message I hear makes it sound like the predictions that come out of these models are certain.
    - bprieto 12y ago
      Thanks. Maybe it's not climate scientist then, but the version of climate science that we get through the media which runs as "science is settled, scientist already know what will be the mean temperature on Earth in 2100".
JoeAltmaier 12y ago
At some point models have to be built from experimental data. I guess the fundamental problem is, we have just one dataset (one planet) so any model is going to be 'overtrained'. Hard to beat that.

by bprieto 12y ago

anon1385 12y ago

>They are saying that the models that predict a global warming catastrophe can't be wrong Who said that? The incompleteness of climate models isn't something that has just come up recently - they have always been incomplete (they can't be anything but - 'all models are wrong') and that is well understood by anybody who develops or uses them. There are large sub-systems that global circulation models don't even really attempt to model at this stage. Others where they are widely recognised to do a very bad job (e.g. ice sheets). Others that are just modelled stochastically because we don't know how to predict those processes on a year to year basis (e.g. ENSO). Finding inconsistencies between the models and reality is one of the ways that they are improved. I don't know where you are getting these ideas about how climate modellers think, but they don't bare any relation to how climate models are actually developed.

bprieto 12y ago
OK, it was a figure of speech. They are not saying that. They are acting like that. The climate scientist who are trying to find the "missing heat" instead of reconsidering their models and revising their predictions of warming.
- anon1385 12y ago
  Why do you think 'find the missing heat' is different to 'reconsidering their models'? If ocean circulation is transporting more heat into the deep ocean than we thought, then that is something to change in the models. If you are asking, why have they not thrown out our entire understanding of the climate system and started over, then I would argue that isn't a way to actually improve our understanding of the climate. We would be starting over every day. Out understanding clearly isn't all wrong. There are all sorts of other lines of evidence used to develop our understanding of the climate system. From things that are reproducible with simple experiments (or just application of basic physical laws), to things that have taken large investments and long term study to work out. There can be all sorts of reasons that the model got it wrong (assuming it did, I don't think we really have enough evidence one way or the other yet for the idea proposed in the OP - it seems like a bit of a curve fitting exercise at the moment). Not only is our understanding of the systems incomplete, our measurements are incomplete as well (and that gets much worse even just a few years ago - we've only had the Argo network of floats taking ocean temperature for a bit over a decade for example). Often there is just as much reason to suspect that it's our measurements that are wrong than that it's our understanding of the system. It's just the nature of working on a system the size of the entire planet. It's not easy to measure everything, and ever harder to measure what was happening in the past. There have been real issues where it was the data that was wrong and not the models. See the UAH troposphere temperature record for example: https://en.wikipedia.org/wiki/UAH_satellite_temperature_dataset#Corrections_made https://en.wikipedia.org/wiki/UAH_satellite_temperature_data... It contradicted other data sets and the models, and it turns out there were errors in the way the temperatures were calculated. That is a record run by scientists very much on the skeptic end of the spectrum by the way, if you were getting worried about any kind of conspiracy.
  - cicero 12y ago
    I think you give some good explanation of why climate modeling is very difficult. However, the mainstream message I hear makes it sound like the predictions that come out of these models are certain.
  - bprieto 12y ago
    Thanks. Maybe it's not climate scientist then, but the version of climate science that we get through the media which runs as "science is settled, scientist already know what will be the mean temperature on Earth in 2100".

JoeAltmaier 12y ago

At some point models have to be built from experimental data. I guess the fundamental problem is, we have just one dataset (one planet) so any model is going to be 'overtrained'. Hard to beat that.